The application of classification tree analysis to soil type prediction in a desert landscape

Scull, Peter; Franklin, Janet; Chadwick, Oliver A.

doi:10.1016/j.ecolmodel.2004.06.036

Cited by 154 publications

(82 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The values for the dark object subtraction were sampled from Fish Lake, Utah, (deep lake) and shadows cast by cumulus clouds. These 5 and 1, 4 and 7, and 3 and 1 exhibited unique patterns wherein distinct landforms and vegetation communities were visually identified and thought to be useful in the model (Cole, 2004;Bodily, 2005;Scull et al, 2005;Nield et al, 2007;Saunders and Boettinger, 2007) (Figure 7, 8, and 9). …”

Section: Landsat-derived Datamentioning

confidence: 99%

“…the conceptual model developed by the soil scientist, then the tree (the set of rules) could potentially predict the soils in unmapped areas. Scull et al (2005) was able to take the next step with decision trees, extracting randomly sampled points from an existing soil survey, to train the trees and then extrapolate into areas that were not previously mapped. Saunders (2005) was able to model soil map units as part of a third order soil survey in Wyoming in a previously unmapped area using classification tree analysis.…”

Section: Soil Spatial Prediction Functionsmentioning

confidence: 99%

“…Specific geomorphic surfaces and positions may approximate relative age (Noller, 2010). Scull et al (2005) represented time implicitly with Landsat imagery which can detect desert varnish.…”

Section: Time (A)mentioning

confidence: 99%

“…Zhu (2000) stratified his study area by geology type before modeling the distribution of soil series using artificial neural networks. Scull et al (2005) stratified their study area into two distinctive physiographic regions, basin and mountain regions, improving the overall accuracy of their prediction with decision tree analysis (DTA).…”

Section: Spatial Position (N)mentioning

confidence: 99%

“…Ongoing research in digital soil mapping has demonstrated that reasonably accurate soil maps can be produced using quantitative predictive models. Digital soil mapping may also expedite soil survey (Lagacherie and Holmes, 1997;Dobos et al, 2000;Zhu, 2000;Moran and Bui, 2002;McBratney et al, 2003;Scull et al, 2003;Cole, 2004;Shi et al, 2004;Henderson et al, 2005;Saunders, 2005;Scull et al, 2005;Cole and Boettinger, 2007;Saunders and Boettinger, 2007;Brungard, 2009).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Stum

2010

Digital Soil Mapping

View full text Add to dashboard Cite

Initial soil surveys are incomplete for large tracts of public land in the western USA. Digital soil mapping offers a quantitative approach as an alternative to traditional soil mapping. I sought to predict soil classes across an arid to semiarid watershed of western Utah by applying random forests (RF) and using environmental covariates derived from Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and digital elevation models (DEM). Random forests are similar to classification and regression trees (CART).However, RF is doubly random. Many (e.g., 500) weak trees are grown (trained) independently because each tree is trained with a new randomly selected bootstrap sample, and a random subset of variables is used to split each node. To train and validate the RF trees, 561 soil descriptions were made in the field. An additional 111 points were added by case-based reasoning using aerial photo interpretation. As RF makes classification decisions from the mode of many independently grown trees, model iii uncertainty can be derived. The overall out of the bag (OOB) error was lower without weighting of classes; weighting increased the overall OOB error and the resulting output did not reflect soil-landscape relationships observed in the field. The final RF model had an OOB error of 55.2% and predicted soils on landforms consistent with soil-landscape relationships. The OOB error for individual classes typically decreased with increasing class size. In addition to the final classification, I determined the second and third most likely classification, model confidence, and the hypothetical extent of individual classes.Pixels that had high possibility of belonging to multiple soil classes were aggregated using a minimum confidence value based on limiting soil features, which is an effective and objective method of determining membership in soil map unit associations and complexes mapped at the 1:24,000 scale. Variables derived from both DEM and Landsat 7 ETM+ sources were important for predicting soil classes based on Gini and standard measures of variable importance and OOB errors from groves grown with exclusively DEM-or Landsat-derived data. Random forests was a powerful predictor of soil classes and produced outputs that facilitated further understanding of soil-landscape relationships.(147 pages) iv ACKNOWLEDGMENTS

show abstract

Section: Landsat-derived Datamentioning

confidence: 99%

Section: Soil Spatial Prediction Functionsmentioning

confidence: 99%

Section: Time (A)mentioning

confidence: 99%

Section: Spatial Position (N)mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Stum

2010

Digital Soil Mapping

View full text Add to dashboard Cite

show abstract

Discrimination of Soils and Assessment of Soil Fertility Using Information from an Ion Selective Electrodes Array and Artificial Neural Networks

Mimendia

Gutiérrez

Alcañiz

et al. 2014

CLEAN Soil Air Water

View full text Add to dashboard Cite

Multichannel sensor measurements combined with advanced treatment is the departure point for a new concept in sensorics, the electronic tongue. Our setup worked with an array of 20 ion selective electrodes plus an artificial neural network used as a pattern recognition method applied to soil analysis. With this design, we got a versatile tool which was able to perform qualitative and quantitative determinations. As first application, the qualitative discrimination between six distinct soil types based on their extractable components was attempted. The procedure was simplified to a single extraction step before measurements. Water, a BaCl 2 saline solution and an acetic acid extract were evaluated as extracting agents. The best performance was reached with the acetic acid extraction method with a correct classification rate and sensitivity both of 94%, and a specificity of 100%. In addition, a quantitative determination of several physicochemical properties of agricultural interest, such as organic carbon content and selected cations (like K þ or Mg 2þ ) and anions (like NO 3 À orCl À ) was also demonstrated, showing satisfactory agreement with the reference methods.Abbreviations: ANN, artificial neural network; ET, electronic tongue; PCA, principal component analysis 1808

show abstract

A case study for evaluating potential soil sensitivity in aridland systems

Peterman¹,

Ferschweiler²

2015

Integr Envir Assess & Manag

View full text Add to dashboard Cite

Globally, ecosystems are subjected to prolonged droughts and extreme heat events, leading to forest die-offs and dominance shifts in vegetation. Some scientists and managers view soil as the main resource to be considered in monitoring ecosystem responses to aridification. As the medium through which precipitation is received, stored, and redistributed for plant use, soil is an important factor in the sensitivity of ecosystems to a drying climate. This study presents a novel approach to evaluating where on a landscape soils may be most sensitive to drying, making them less resilient to disturbance, and where potential future vegetation changes could lead to such disturbance. The drying and devegetation of arid lands can increase wind erosion, contributing to aerosol and dust emissions. This has implications for air quality, human health, and water resources. This approach combines soil data with vegetation simulations, projecting future vegetation change, to create maps of potential areas of concern for soil sensitivity and dust production in a drying climate. Consistent with recent observations, the projections show shifts from grasslands and woodlands to shrublands in much of the southwestern region. An increase in forested area occurs, but shifts in the dominant types and spatial distribution of the forests also are seen. A net increase in desert ecosystems in the region and some changes in alpine and tundra ecosystems are seen. Approximately 124,000 km(2) of soils flagged as "sensitive" are projected to have vegetation change between 2041 and 2050, and 82,927 km(2) of soils may become sensitive because of future vegetation changes. These maps give managers a way to visualize and identify where soils and vegetation should be investigated and monitored for degradation in a drying climate, so restoration and mitigation strategies can be focused in these areas.

show abstract

The application of classification tree analysis to soil type prediction in a desert landscape

Cited by 154 publications

References 22 publications

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Random Forests Applied as a Soil Spatial Predictive Model in Arid Utah

Discrimination of Soils and Assessment of Soil Fertility Using Information from an Ion Selective Electrodes Array and Artificial Neural Networks

A case study for evaluating potential soil sensitivity in aridland systems

Contact Info

Product

Resources

About